Previously, we didn't retry request on connection failure. Sometimes
we hit the edge case where connection is about to lost just when we
write request. To avoid this situation, we now retry request to
failed attempt. We also add ConnectBlocker to MemcachedConnection not
to attempt to connect to memcached if connection could not be made
previously.
Some API processing is very slow (e.g., getaddrinfo). To avoid to
slow down regular request handling, if multi threaded configuration is
enabled, we allocate dedicated worker for API.
To keep ipc channel being read from worker process, default loop
should not be stopped. To join all worker threads, now we use
dedicated thread to do this. When all worker threads are joined,
ev_async_send sends message to default loop, and it is finally
stopped.
Since libev handles SIGCHLD, using waitpid in separate thread to wait
for the completion of fetch-ocsp-response script process is undefined.
This commit rewrite ocsp handling code so that it utilizes libev
ev_child watcher and perform ocsp update without thread.
When same SSL_CTX is used by multiple thread simultaneously we have to
setup some number of mutex locks for it. We could not check how this
locking affects scalability since we have 4 cores at best in our
development machine. Good side of sharing SSL_CTX across threads is
we can share session ID pool.
If --tls-ctx-per-worker is enabled, SSL_CTX is created per thread
basis and we can eliminate mutex locks. The downside is session ID is
no longer shared, which means if session ID generated by one thread
cannot be acceptable by another thread. But we have now session
ticket enabled and its keys are shared by all threads.
Tatsuhiro Tsujikawa
Tomasz Buchert